ARTICLE IN PRESS
Journal of Crystal Growth 310 (2008) 1691–1696 www.elsevier.com/locate/jcrysgro
Phase, nucleation and coalescence of HgI2 onto amorphous substrates Ana Lı´ a Noguera, Marı´ a Eugenia Pe´rez, Eduardo Quagliata, Laura Rosa Fornaro Compound Semiconductors Group, Faculty of Chemistry, University of the Republic—Montevideo, Gral. Flores 2124, Montevideo, Uruguay Available online 14 December 2007
Abstract Heterogeneous nucleation and further growth of HgI2 onto glass were studied by the physical vapor deposition (PVD) method, with and without an argon atmosphere. Glass substrates 200 200 in area were used for nucleation and growth. Supersaturation strongly determined not only the nuclei population and size, but also the HgI2 nucleation phase. A range of nucleation–growth temperatures between 303 and 333 K for the substrate, and a range of initial argon pressure between 1.3 and 2.6 104 Pa were found appropriate for nucleation and growth in the a (red) phase of HgI2. Other conditions, such as an argon pressure higher than 4 104 Pa, determined the nucleation and growth in the b (yellow) metastable phase. After a nucleation growth time of 30 min, clusters of about 0.5 mm for b-HgI2 and 1 mm for a-HgI2 and a minimum distance between the clusters of about 1 and 10 mm were, respectively, obtained. Coalescence was performed by annealing the clusters at a temperature between 303 and 333 K and from 5 min to 24 h, at an initial argon pressure of 104 Pa. They coalesced to form larger clusters, which, in the case of the b-phase also transformed to the a-stable phase. From the experimental nucleation–growth and further coalescence conditions, correlations between supersaturation conditions and phase were established. Also, considerations about interface free energies and nucleation phase were made. Future work will be conducted to diminish nuclei size and enlarge nuclei population, seeking for early stages of nucleation. r 2007 Elsevier B.V. All rights reserved. PACS: 68.55.A; 68.35.R1; 81.15.Kk Keywords: A1. Nucleation; A3. Graphoepitaxy; A3. Physical vapor deposition processes; B2. Semiconducting mercury compounds
1. Introduction HgI2 crystalline layers have been studied for more than 10 years as photoconductors for direct and digital X-ray imagers [1,2]. These devices are based on the growth of layers of materials appropriate for X-ray detection onto a readout-active matrix such as a thin-film transistor (TFT). In these devices, each pixel of the array will be a rear contact, which will give the signal for the corresponding pixel in the final image. A metallic thin film, deposited on top of the semiconductor, will act as front electrode. When radiation is absorbed, electron–hole pairs are produced in the photoconductor, and the applied field will direct them towards the electrodes [3]. HgI2 has very suitable properties for X-ray detection, such as a high atomic number, a high density, an adequate Corresponding author.
E-mail address:
[email protected] (L.R. Fornaro). 0022-0248/$ - see front matter r 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2007.12.005
bandgap at room temperature and high-radiation absorption coefficient for the X-rays used in most of imaging applications. For this reason, it is one of the most promising materials for direct radiography [1]. Although mercuric iodide layers have been grown by several methods, physical vapor deposition (PVD) has been the most employed, and has given polycrystalline but oriented layers as well [1,2,4]. If reported results are reviewed, it is clear that the more oriented the film, and the higher the crystalline quality of the film, the better its electrical and response-to-radiation properties [5]. These results confirm theoretical considerations concerning the charge transport through a layer as long as their collection at the electrodes agree with the results reported for layers of other semiconductors used for a variety of applications. The trend on improving mercuric iodide layers orientation and consequently, X-ray imagers performance, indicates that the main challenge of the development of direct X-ray imagers made with them is to grow the most
ARTICLE IN PRESS 1692
A.L Noguera et al. / Journal of Crystal Growth 310 (2008) 1691–1696
oriented, or, even better still, epitaxial layers onto the TFT matrix, which is an amorphous -and non-homogeneoussubstrate [5]. This growth requires, first of all a good understanding of the first steps of graphoepitaxy, which means, of the nucleation and coalescence of the crystalline material onto amorphous substrates. However, there are no reports, neither experimental nor theoretical, about nucleation or coalescence intended for growing layers onto amorphous substrates, neither of mercuric iodide nor of the other compounds of the heavy metal halides family. On the other hand, it is well known that mercuric iodide nucleates and crystallizes in several phases, being the most studied the a-HgI2 (red, tetragonal) and the b-HgI2 (yellow, orthorhombic) phases, with a transition temperature at about 400 K [6]. Consequently, the occurrence of phase transitions during nucleation and/or coalescence of mercuric iodide layers may be reasonably expected. In the light of these antecedents, this work reports the first approach to the study of heterogeneous nucleation and coalescence of mercuric iodide onto amorphous substrates intended for further layer growth, as long as the study of the nuclei phase and the phase transitions concurrently involved in those phenomena.
2. Methods Mercuric iodide Aldrich 99% purified by four sublimations, at 493 K and at an initial pressure of 103 Pa was used as a starting material. Nucleation and further growth were performed by the PVD method onto 200 200 glass substrates. The system used was specially designed and constructed to get a close control of parameters such as geometrical disposition of the starting material, material source temperature and deposition temperature and time, within certain constraints. Conditions were established for studying mercuric iodide nucleation and further growth as a function of supersaturation. Supersaturation was estimated as Dm ¼ (mvmc), where mv and mc are the chemical potentials of vapor and crystal, respectively. These conditions implied source temperatures between 293 and 373 K, deposition temperatures between 293 and 373 K, deposition times between 30 and 1800 s and initial pressures from 103 to 4 104 Pa with and without Ar inert atmosphere. Deposition temperatures were set with an uncertainty of about 71 K. Chemical potentials were estimated according to m ¼ RT ln P, with T being the absolute mercuric iodide source and crystal temperatures, and P the vapor pressure at those temperatures, estimated according to log P ¼ 30.27–6.47 log T–5690 T1 (with P in Torr) [7]. Coalescence was performed annealing at temperatures between 303 and 333 K from 5 min to 24 h at an initial argon pressure of 104 Pa. Nucleation and coalescence were characterized by optical microscopy using a Nikon Model EPIPHOT 300
microscope. Mean nuclei size and height were estimated by optical microscopy. 3. Results and discussion The study of mercuric iodide nucleation and growth on amorphous substrates was performed as a function of supersaturation. The vapor pressure of mercuric iodide determines high supersaturation and fast initial growth even for low source–substrate temperature differences. In order to seek for the first steps of growth, a low supersaturation was established. On the other hand, let us consider the free energy for a prismatic (square shape) nucleus created by heterogeneous nucleation, which follows the equation [8–10]: Dg ¼ l 2 hDm þ l 2 ðgnv þ gsn þ gsv Þ þ 4lhgnv
(1)
where l is the length of the nucleus, h the height of the nucleus, Dm the supersaturation, gnv the specific free energy of the nucleus–vapor interface, gsn the specific free energy of the substrate–nucleus interface, and gsv the specific free energy of the substrate–vapor interface. This free Gibbs energy has two components: one directly proportional to supersaturation and the other which depends on the interface energies. The interface energies can make negative the free Gibbs energy allowing nucleation for positive, but also for zero and even negative supersaturation conditions. In the light of the previous considerations, we studied nucleation and further growth of mercuric iodide onto amorphous substrates by establishing low-supersaturation conditions, and obtained the results summarized in Table 1. Before discussing and taking any conclusion from the obtained experimental results, some warnings should be considered. From a theoretical point of view, the Gibbs–Thomson relation indicates that, the lower the supersaturation, the larger the nucleus size. Therefore, nucleation at low supersaturation should produce quite large nuclei. From an experimental point of view, reports referring to nucleation of other materials indicate that nuclei size as long as the distance between the nuclei are in the order of nanometers [11,12]. Taking into account the experimental constraints, our nucleation–growth system permits to obtain a minimum nuclei size of about 0.5 mm Table 1 Dm
Source temperature range (T/Tm)
40 0.59–0.70 0 0.57–0.63 0.63
o0 0.57–0.59
Substrate temperature range (T/Tm)
Initial pressure range (Pa)
Phase of mercuric iodide nuclei
0.57–0.67
104–104
b-HgI2 (yellow)
0.57–0.63 0.63
0.58–0.61
4
3.9 104
a-HgI2 (red) a-HgI2 (red) a-HgI2 (red) and b-HgI2 (yellow) b-HgI2 (yellow)
104
No nucleation
10 1.3 104 2.6 104
ARTICLE IN PRESS A.L Noguera et al. / Journal of Crystal Growth 310 (2008) 1691–1696
for b-HgI2 and about 1 mm for a-HgI2, and a minimum distance between nuclei of about 1 and 10 mm, respectively. Previous reports about mercuric iodide nucleation for bulk crystal growth from the vapor do not indicate nuclei size, but nuclei population density, which was in the order of 3–55 nuclei/cm2 [6]. Our values of nuclei sizes lead us to think that a first nucleation in the nanometer ranged size and further coalescence may have been taken place during what we named ‘‘nucleation time’’. If this would have been the case, we are not observing ‘‘nuclei’’, but larger clusters due to nucleation and further growth. The following discussion should be considered in the light of this, and conclusions will have to be completed by further studies in the nanometer scale. First of all, for mercuric iodide heterogeneous nucleation, Table 1 indicates that a negative supersaturation does not give any nuclei. Nucleation with non-positive supersaturation has never been reported for mercuric iodide or other heavy metal halides; however, it has been already experimentally achieved for other compounds [9]. A remarkable result of Table 1 is that the nucleation of mercuric iodide can take place close to zero supersaturation (within the experimental uncertainty). Even more, proximity to zero supersaturation and a range of initial pressure between 104 and 1.3 104 Pa of argon are required for sole nucleation in the a-HgI2 desirable phase, stable at room temperature. Other nucleation conditions, such as a higher-temperature gradient between source and substrate or an argon pressure higher than 2.6 104 Pa, determine the nucleation in the b-HgI2 metastable phase, or no nucleation. Fig. 1 shows representative nucleation and further growth in the a-HgI2 phase. Clusters are red, and clearly square shaped, in agreement with the tetragonal structure of the a-HgI2 phase. From a series of microscopic images as a function of the deposition temperature, mean cluster sizes were estimated and gave the Arrhenius dependence that can be seen in Fig. 2. This Arrhenius plot, obtained from clusters in the micrometer scale, corresponds to growth stages of mercuric iodide clusters. Previous kinetic studies about the growth of mercuric iodide crystals determined a constant growth rate for crystals up to about 7 mm in size [6]. Considering this condition of constant
1693
growth rate, that the growth time for all the clusters was the same (30 min), and that the Arrhenius plot does not show any slope variation, we can assume that the cluster size is proportional to the growth rate. Then, from Fig. 2, an activation energy for growth, Eg, 0.3370.03 eV can be estimated for the a-HgI2 phase. In our knowledge, this is the first time that an activation energy for the growth of mercuric iodide is reported; therefore, we cannot compare this value with previous reported ones. However, it is in the order of other activation energies for growth, for instance, of GaAs, for which a value of 0.73 eV was reported [13]. The limit conditions achieved to obtain the red phase, together with the smallest cluster size and distance between the clusters, were close to zero supersaturation with a source temperature of T/Tm ¼ 0.63 and a substrate temperature of T/Tm ¼ 0.63, as long as an argon initial pressure of 1.3 104 Pa. Fig. 3 shows representative results for close to zero supersaturation where both a-HgI2 and b-HgI2 phases were obtained. Crystallites of the both phases co-exist in the same layer, each phase with its structural properties. After a period of time—which depends on temperature and light exposition—b-HgI2 crystallites transform to a-HgI2 ones, giving stressed and dendrite-shaped clusters. Thermodynamics predicts that the further growth at zero supersaturation is not possible; this was experimentally
Fig. 2. Arrhenius plot: natural logarithm of mean cluster size (mean cluster size in microns) as a function of 1/T (K1) for the a-HgI2 phase (initial pressure 104 Pa).
Fig. 1. Representative nucleation and further growth in the a-HgI2 phase (Dmffi0) (deposition temperature: (a) T/Tm ¼ 0.57 and (b) T/Tm ¼ 0.63).
ARTICLE IN PRESS 1694
A.L Noguera et al. / Journal of Crystal Growth 310 (2008) 1691–1696
Fig. 3. Representative nucleation and further growth in both a-HgI2 and b-HgI2 phases (Dmffi0) (deposition temperature: T/Tm ¼ 0.63).
Fig. 4. Representative nucleation in the b-HgI2 phase (Dm40) (deposition temperature T/Tm ¼ 0.61).
observed by increasing deposition time with no change in the cluster size and the cluster population density. On the other hand, Table 1 indicates that a positive supersaturation always gives b-HgI2 phase nucleation. Fig. 4 shows a representative nucleation and further growth in the b-HgI2 (yellow) phase. Clusters are yellow and clearly rhombohedral in agreement with the orthorhombic structure of the b-HgI2 phase. From a series of microscopic images as a function of the deposition nucleation temperature, mean cluster sizes were estimated and gave the Arrhenius dependence that can be seen in Fig. 5. As for Fig. 2, this Arrhenius plot obtained from clusters in the micrometer scale corresponds to the growth stages of mercuric iodide clusters. With the same considerations made for Fig. 2, an activation energy for growth, Eg, 0.4570.04 eV can be estimated from Fig. 5 for the b-HgI2 phase. Again, this value cannot be compared
Fig. 5. Arrhenius plot: natural logarithm of mean cluster size (mean cluster size in microns) as a function of 1/T (K1) for the b-HgI2 phase (initial pressure 104 Pa).
with previous ones for mercuric iodide, but is in the order of values obtained for other materials. Another interesting issue from Table 1 and Fig. 4 is that, although the yellow phase should be stable above 400 K, b-HgI2 nuclei and growth were obtained at far below the transition temperature. The existence of the metastable yellow phase within the stability range of the red one was already reported years ago [14,15]. That means that the nucleation and further growth of mercuric iodide completely follow the Ostwald rule: the metastable yellow phase may appear exclusively or concurrently with the stable red phase [6]. Before discussing coalescence results, similar warnings than for nucleation ones have to be considered. That means: coalescence results reported for other materials involve nanometer-sized nuclei [11,12], but our ‘‘coalescence’’ experiments involve micrometer-sized ‘‘nuclei’’. Although oversized, and with long distances between clusters, the first obtained a-HgI2 clusters coalesce, as can be seen in Fig. 6. Clusters tend to lose their square shape and to agglutinate in the same layer as long as they seem to maintain their initial growth orientation. From Fig. 6, it is clear that the distance between clusters is larger and that cluster population is lower than necessary for a complete coverage of the substrate. Coalescence results for b-HgI2 are similar to the ones for a-HgI2 as long as they involve too large and separate clusters, and lack of a complete substrate coverage. However, as it can be seen in Fig. 7, coalescence not only implies cluster aggregation but also heat and time lead to the transition to the a-HgI2 room-temperature stable phase. The cluster size and shape dramatically change with coalescence: size increases (via agglutination of b-HgI2 crystallites) and shape abruptly changes, from yellow rhombohedral to red and larger clusters, composed by several square-shaped (of tetragonal a-HgI2) grains.
ARTICLE IN PRESS A.L Noguera et al. / Journal of Crystal Growth 310 (2008) 1691–1696
1695
Fig. 6. Coalescence of a-HgI2 clusters obtained after annealing at 313 K during 24 h.
Fig. 7. Coalescence of b-HgI2 obtained after annealing at 313 K during 24 h.
The obtained results have to be taken as a first approach to the phenomena. That is, we have found conditions to separately nucleate and grow mercuric iodide in two of its phases. It is clear now that, for studying nucleation in the appropriate range of nuclei size, at nanometer scale, our experimental constraints must be overcome. Especially, lower nucleation times and as — Figs. 2 and 5 indicate — lower nucleation temperatures will have to be experimentally achieved. Our results for this case of heterogeneous nucleation and growth on amorphous substrates, with two solid phases nucleating under different supersaturation conditions, lead us to think that the interface energies of Eq. (1) — different for each mercuric iodide phase, although not too much taking into account the similar atomic cell parameters — might be playing an important role in determining the phase of nuclei. The above-mentioned study of nucleation in the nanometer scale should also permit a deep understanding of the influence of interface energies on the phase of nuclei.
necessary condition for nucleation in the a-HgI2 phase. Results lead to think that the interface free energies play a crucial role in determining the nuclei phase, looking for the most favorable free Gibbs energy for nuclei formation. It must be taken into account that these conclusions proceed from studying larger than appropriate nuclei and distance between nuclei — both determined by experimental constraints — and must be then considered as the first micrometer-scale approach to the oriented crystallization of mercuric iodide on amorphous substrates. Future work will be conducted to diminish the nuclei size, enlarge the nuclei population and consequently diminish the distance between nuclei, seeking for a more accurate study of nucleation and coalescence of the material. Furthermore, nuclei orientation will have to be determined by X-ray studies, for rigorously assigning free interface energies ðgh k l Þ to the correspondent crystal planes.
4. Conclusions
We thank H. Bentos Pereira for helping with the growth systems, and PEDECIBA, CSIC and PDT Uruguayan Programs for financial support.
Nucleation and coalescence results for a-HgI2 and b-HgI2 onto amorphous substrates, intended for further film growth, are here reported for the first time. They indicate that mercuric iodide nucleation phase depends on supersaturation conditions, with the proximity to zero supersaturation — between certain pressure limits — as the
Acknowledgments
References [1] L. Fornaro, A. Cun˜a, A. Noguera, I. Aguiar, M. Pe´rez, L. Mussio, A. Gancharov, IEEE Trans. Nucl. Sci. 52/6 (2006) 3107.
ARTICLE IN PRESS 1696
A.L Noguera et al. / Journal of Crystal Growth 310 (2008) 1691–1696
[2] G. Zentai, L. Partain, R. Pavlyuchkova, C. Proano, G. Virshup, L. Melekhov, A. Zuck, B. Breen, O. Dagan, A. Vilensky, M. Schieber, H. Gilboa, P. Bennet, K. Shah, Y. Dmitriyev, J. Thomas, M. Yaffe, D. Hunter, Proc. SPIE 5030 (2003) 77. [3] L.E. Antonuk, Phys. Med. Biol. 47 (2002) R31. [4] J.S. Iwanczyk, B.E. Patt, C.R. Tull, L.R. MacDonald, N. Skinner, E.J. Hoffman, L. Fornaro, L. Mussio, E. Saucedo, A. Gancharov, Proc. SPIE 4508 (2001) 28. [5] L. Fornaro, I. Aguiar, A. Noguera, M. Pe´rez, N. Sasen, L. Mussio, IEEE Nucl. Sci. Symp. Conf. Rec. 7 (2005) 878. [6] M. Piechotka, Mater. Sci. Eng. R18 (1–2) (1997) 1. [7] K. Conder, J. Laskowski, Nucl. Instrum. Methods—Phys. Res. A 283 (1989) 138. [8] A.A. Chernov, Modern Crystallography III, Crystal Growth, Springer, Berlin, Heidelberg, New York, Tokyo, 1984, p. 66. [9] I. Givargizov, Oriented Crystallization on Amorphous Substrates, first ed., Plenum Press, New York, 1991, p. 6.
[10] P. Rudolph, R. Fornari, C. Paorici (Eds.), Theoretical and Technological Aspects of Crystal Growth, Trans. Tech. Publ., Switzerland, 1998, pp. 1–26. [11] J.W. Evans, P.A. Thiel, M. Li, in: M. Slowronski, J.J. De Yoreo, C.A. Wang (Eds.), Perspectives on Inorganic, Organic, and Biological Crystal Growth: From Fundamentals to Applications, American Institute of Physics, USA, 2007, pp. 191, 206. [12] M. Giesen, in: M. Slowronski, J.J. De Yoreo, C.A. Wang (Eds.), Perspectives on Inorganic, Organic, and Biological Crystal Growth: From Fundamentals to Applications, American Institute of Physics, USA, 2007, pp. 115, 124. [13] M. Borgsto¨m, K. Deppert, L. Samuelson, W. Seifert, J. Crystal Growth 260 (2004) 18. [14] N.V. Long, K. Kleinstu¨ck, J. Tobisch, P. Linger, K. Prokert, V. Schricht, Cryst. Res. Technol. 18 (1983) K93. [15] G. Perrier, C. Belouet, J. Omaly, R. Cadoret, in: M.J. Rucroft (Ed.), COSPAR Space Research, vol. 29, Pergamon Press, Oxford, 1979, p. 531.